Abstract
Dynamic peer behavior in P2P networks has a large impact on the performance because the peer acts simultaneously as the server and client of an overlay network. However, peer behavior in pure P2P networks is not well known because there is no management system and the size is large. We have proposed a measurement method to efficiently collect information of an entire network for Winny, which is the most popular P2P file sharing application in Japan. Especially, we analyze behaviors of the peer with/without files and identify long life peers in the network. The analysis results show that the network is composed of about 30 % of observed peers in 24 hours, and 50 % of peers contribute to the network as the files uploaders. Selfish peers, which leave the network after downloading, are observed especially in the morning. The file uploading peers and the joining peers after the midnight live long period in the network.